Rodrigo Castellano
AI Researcher

Rodrigo Castellano Ontiveros

Marie Skłodowska-Curie Fellow, Learning with Multiple Represenations (LeMuR) project.

My research aims to make Neural-Symbolic reasoning both scalable and transparent. I use Reinforcement Learning methods for efficiency, and for explainability, I work on interpretable-by-design models that operate on first-order logic.

I'm excited to contribute to projects on AI reasoning, probabilistic programming, and scalable symbolic/NeSy AI. Other interesting domain is LLM agents and tool use, focusing on how to embed logical reasoning capabilities within or alongside LLMs. I’m also open to collaborating on new and adjacent topics where I can learn and apply my expertise.

Contact Me Scholar

Where I Have Been

My academic and professional journey across Europe.

Research Papers

AAAI 2026
DeepProofLog: Efficient Proving in Deep Stochastic Logic Programs

Y. Jiao*, R. Castellano Ontiveros*, L. De Raedt, M. Gori, F. Giannini, M. Diligenti, G. Marra

A novel NeSy system mapping the resolution process of Deep Stochastic Logic Programs to Markov Decision Processes, enabling efficient Reinforcement Learning for logical proving.
* Equal contribution

Paper Code
IJCAI 2025
Grounding Methods for Neural-Symbolic AI

R. Castellano Ontiveros, F. Giannini, M. Gori, G. Marra, M. Diligenti

Proposes a parameterized family of grounding methods generalizing Backward Chaining to control the trade-off between scalability and expressiveness.

Paper Code
XAI 2025
Interpretable Link Prediction via Neural-Symbolic Reasoning

R. Castellano Ontiveros, E. Bonabi Mobaraki, F. Giannini, P. Barbiero, et al.

Introduces interpretable-by-design R-CBM-style models that output explicit proof trees, evaluated using XAI metrics like coherence.

Paper Code
NeSy 2025
Distilling KGE Black Boxes into Interpretable NeSy Models

R. Castellano Ontiveros, F. Giannini, M. Diligenti

A framework to distill Knowledge Graph Embeddings into interpretable Neural-Symbolic models, ensuring high fidelity while providing logic proofs.

Paper Code
Nature
A Machine Learning-based Approach for Constructing Remote Photoplethysmogram Signals

R. Castellano Ontiveros, M. Elgendi, C. Menon

Developed a novel ML methodology that achieved SOTA performance in rPPG signal reconstruction from video, allowing the extraction of different physiological parameters (vs. single-feature prediction competitors).

Paper Code
J. Cheminf.
Physicochemical modelling of the retention mechanism of temperature-responsive polymeric columns

Elena Bandini, Rodrigo Castellano Ontiveros, Ardiana Kajtazi, Hamed Eghbali, Frédéric Lynen

Applied machine learning algorithms to model the retention mechanisms in HPLC columns.

Paper
F. Physiol.
Evaluating RGB channels in remote photoplethysmography: a comparative study

Rodrigo Castellano Ontiveros, Mohamed Elgendi, Giuseppe Missale, Carlo Menon

A comparative study of the efficacy of RGB channels in remote photoplethysmography (rPPG) when compared with contact-based PPG.

Paper

Software & Projects

A collection of open-source tools and implementations focusing on Neural-Symbolic AI, Computer Vision, and others.

Desion Preview

Camera Inspection System

Automated visual inspection systems using Computer Vision and synthetic data created with Unreal Engine.

Computer Vision • Industry

NeSy Grounding

Parametrized grounding methods for scalable neural-symbolic reasoning.

Neuro-symbolic AI • First Order Logic

rPPG Construction

Physiological parameters extraction from video using Machine Learning techniques.

Deep Learning

Cosmic Rays Analysis

Prediction of particles associated to cosmic rays detected by Cherenkov Telescope Array.

Data Science

Accent Embeddings

Speech technology project focused on accent classification.

Audio Processing

Experience

PhD Researcher in AI

Nov 2023 — Present
University of Siena, Italy
Marie Skłodowska-Curie Doctoral Fellowship

MSCA Project LeMuR: Learning with Multiple Representations.

Supervisors: Prof. Marco Gori, Prof. Michelangelo Diligenti, Dr. Francesco Giannini.

Visiting Researcher

Oct 2025 — Present
Baker Hughes, Florence

Integrating logic reasoning systems with LLM agents.

Visiting Researcher

Sep 2024 — Dec 2024
KU Leuven, Belgium

Developed DeepProofLog. Collaboration with Ying Jiao, Prof. Giuseppe Marra, and Prof. Luc De Raedt.

International Talent Programme Trainee

Apr 2023 — Oct 2023
ING Bank, Brussels

Participated in a rotational data science/analytics graduate program.

Master Thesis Researcher

Oct 2022 — Feb 2023
ETH Zürich, Switzerland

rPPG: extraction from video. [Nature Portfolio] [Frontiers]

Awarded Karl Engver's Foundation Grant.

Supervised by Prof. Carlo Menon and Prof. Moe Elgendi.

Teaching Assistant

Nov 2021 — Apr 2022
KTH Royal Institute of Technology, Sweden
Supported courses in Artificial Intelligence (DD2380) and Machine Learning (DD2421).
Desion Preview
Oct 2020 — Jun 2021

AI Applied Scientist

Desion (Fraunhofer IGD Spin-off), Germany

Built automated visual inspection systems with Computer Vision models (Mask R-CNN) using synthetic data. Synthetic data created with Unreal Engine. [See Demo]

Education

MSc in Machine Learning

KTH Royal Institute of Technology, Stockholm

Includes exchange semester at RWTH Aachen

Apr 2021 — Apr 2023

Exchange Bachelor Student

University of Helsinki, Finland
Sept 2018 — Jun 2019

BSc in Physics

University of Granada, Spain
Oct 2015 — Nov 2020